Reference
Aviation compliance and AI glossary
The terms the Insights guides rely on, defined in plain language. Each entry names the regulation, standard or guide it comes from; the guides quote the rule text itself. Definitions are ours, not the regulator's, and were checked against the current EASA, FAA and JARUS texts on 2026-09-17.
Every primary document cited on this site is listed in the sources register.
- Certification basis
The set of airworthiness and environmental requirements that a product must meet to receive a type certificate. Under EASA it is established by the Agency under point 21.B.80 from the certification specifications applicable at the date of application, plus any special conditions, equivalent safety findings and elected later amendments. The FAA establishes it under 14 CFR 21.17 with the same building blocks. Every row of a compliance matrix traces back to one requirement in it.
Source: Regulation (EU) No 748/2012, point 21.B.80; 14 CFR 21.17 · Used in Certification-basis compliance matrix: EASA Part 21 and FAA
- Compliance matrix (compliance checklist)
The table that lists every requirement of the certification basis, states whether and how it applies to the design, names the means of compliance and points to the compliance document that carries the evidence. EASA calls it the compliance checklist and expects it in the certification programme under AMC 21.A.15(b). The FAA expects the same table in the certification plan under Order 8110.4C. It is the document a certification project is managed from.
Source: AMC 21.A.15(b) Content of the certification programme; FAA Order 8110.4C · Used in Certification-basis compliance matrix: EASA Part 21 and FAA
- Means of compliance (MC0 to MC9)
The method by which compliance with one requirement is demonstrated. EASA codes them in Appendix A to AMC 21.A.15(b): MC0 compliance statement and design data, MC1 design review, MC2 calculation or analysis, MC3 safety assessment, MC4 laboratory tests, MC5 ground tests, MC6 flight tests, MC7 design inspection or audit, MC8 simulation, MC9 equipment qualification. The code chosen for a row decides which compliance document has to exist for it.
Source: Appendix A to AMC 21.A.15(b) Means of compliance codes · Used in Certification-basis compliance matrix: EASA Part 21 and FAA
- Special condition
An additional airworthiness requirement prescribed by the authority when the certification specifications do not contain adequate or appropriate safety standards for a product, typically because of a novel or unusual design feature. EASA issues them under point 21.B.75, the FAA under 14 CFR 21.16. Once issued, a special condition is part of the certification basis and needs its own rows in the compliance matrix. SC-VTOL and SC Light-UAS are examples that became de facto standards for new categories.
Source: Regulation (EU) No 748/2012, point 21.B.75; 14 CFR 21.16 · Used in Certification-basis compliance matrix: EASA Part 21 and FAA
- AMC and GM (acceptable means of compliance, guidance material)
Non-binding material issued by EASA alongside the binding rules. An acceptable means of compliance describes one way of meeting a rule that the Agency accepts; an applicant may propose an alternative means of compliance instead. Guidance material explains a rule or an AMC without adding requirements. Keeping the three levels apart, rule, AMC and GM, is one of the places a general-purpose assistant most often goes wrong.
Source: Regulation (EU) 2018/1139, Article 76(3) and Article 115 · Used in Can AI be trusted for aviation compliance documentation?
- Instructions for continued airworthiness (ICA)
The maintenance, inspection and repair information a type certificate holder must produce so that the product can be kept airworthy in service. EASA requires them under point 21.A.7, the FAA under 14 CFR 21.50. They feed the maintenance data that a Part 145 organisation must hold current under point 145.A.45 and that a repair station must keep current and accessible under 14 CFR 145.109(d).
Source: Regulation (EU) No 748/2012, point 21.A.7; 14 CFR 21.50 · Used in AI assistant for EASA Part 145 and CAMO organisations
- Maintenance organisation exposition (MOE)
The document in which an approved maintenance organisation describes itself to its authority: scope of work, facilities, personnel, procedures and the management system. Point 145.A.70 prescribes its content and the AMC prescribes its layout. Because it and the procedures it incorporates by reference are already the controlled description of how the organisation works, the MOE is the natural corpus for an AI assistant in a Part 145 organisation.
Source: Regulation (EU) No 1321/2014, point 145.A.70 · Used in AI assistant for EASA Part 145 and CAMO organisations
- Continuing airworthiness management exposition (CAME)
The equivalent of the MOE for a continuing airworthiness management organisation, prescribed by point CAMO.A.300. It describes how the CAMO manages the airworthiness of the aircraft under its responsibility, including the maintenance programme, airworthiness review and the contracts with maintenance organisations. Together with the MOE it defines what an assistant for a combined organisation is allowed to answer from.
Source: Regulation (EU) No 1321/2014, point CAMO.A.300 · Used in AI assistant for EASA Part 145 and CAMO organisations
- Maintenance data
The applicable instructions a maintenance organisation must hold and use when carrying out maintenance: the requirements and procedures of the competent authority, airworthiness directives, instructions for continued airworthiness, manufacturer data and approved repair data. Point 145.A.45 makes the organisation responsible for holding it current and keeping the data it controls up to date. 14 CFR 145.109(d) places the same obligation on a certificated repair station. An assistant that cannot show which revision it answered from does not help the organisation meet either point.
Source: Regulation (EU) No 1321/2014, point 145.A.45; 14 CFR 145.109(d) · Used in Wingman360 vs ChatGPT, Claude and Gemini for aviation work
- Certificate of release to service (CRS)
The certificate issued by authorised certifying staff, on behalf of the maintenance organisation, once they have verified that all ordered maintenance was properly carried out in accordance with the exposition procedures and the applicable maintenance data. Point 145.A.50 governs it. It is one of the acts an AI assistant never performs: it can find the procedure and the data, but the verification and the signature are the certifying staff's.
Source: Regulation (EU) No 1321/2014, point 145.A.50 · Used in AI assistant for EASA Part 145 and CAMO organisations
- Nominated person
A person named in the exposition who is responsible to the accountable manager for a defined area, such as maintenance, quality or continuing airworthiness management, and who must be accepted by the competent authority. Nominated persons hold the responsibilities that stay with people when an AI assistant is introduced: approving procedures, deciding classifications, reviewing and signing what the assistant drafted.
Source: Regulation (EU) No 1321/2014, point 145.A.30 and point CAMO.A.305 · Used in AI assistant for EASA Part 145 and CAMO organisations
- Part-IS (information security)
The EASA rules requiring aviation organisations and authorities to manage information security risks with a potential impact on aviation safety. Regulation (EU) 2022/1645 applies to design and production organisations, Regulation (EU) 2023/203 to maintenance organisations, CAMOs, operators, training and ATM organisations. Point IS.I.OR.200 requires an information security management system; IS.I.OR.205 the identification and assessment of risks, including interfaces with other organisations; IS.I.OR.235 the management of contracted activities. An AI assistant that reads controlled documents falls inside all three.
Source: Easy Access Rules for Information Security, points IS.I.OR.200, 205 and 235 · Used in On-premise and air-gapped AI for aviation and defence
- Information security management system (ISMS)
The management system Part-IS requires: a policy on information security, the identification and review of risks, risk treatment measures, an internal reporting scheme, and the measures that keep the system effective. It is the framework within which the decision to adopt any AI tool, cloud or on-premise, has to be recorded. The inventory of systems, data and interfaces it requires is what turns "where does the assistant run" from a preference into an assessed decision.
Source: Easy Access Rules for Information Security, point IS.I.OR.200 · Used in On-premise and air-gapped AI for aviation and defence
- Air-gapped deployment
An installation with no connection to the internet or to any external network. For an AI assistant this means the language model itself runs on hardware inside the organisation, because there is no path to a cloud model endpoint. It is a different product from a web application with a private label, and it is the deployment defence suppliers and organisations handling data under contract usually require.
Source: Wingman360 Teammate deployment models · Used in On-premise and air-gapped AI for aviation and defence
- Single-tenant deployment
A dedicated instance of the software, its database and its document store provisioned for one organisation, in that organisation's own cloud account or on its own servers, sharing nothing with any other customer. It is the opposite of a multi-tenant service where many customers share one installation. Under Part-IS it keeps the interface inventory short: the only external interface is the model endpoint, and with a local model there is none.
Source: Wingman360 Teammate deployment models · Used in On-premise and air-gapped AI for aviation and defence
- Functional hazard assessment (FHA)
A systematic, top-down evaluation of the functions of an aircraft or a system that identifies the failure conditions of each function, determines their effects by flight phase and operating condition, classifies the severity of each effect, and records the assumptions and rationale. It is done at aircraft level (AFHA) and system level (SFHA) under ED-135 and SAE ARP4761A, both of December 2023, and its classifications set the safety objectives the design must meet. ED-279 adapts it for unmanned aircraft.
Source: EUROCAE ED-135 / SAE ARP4761A (2023); EUROCAE ED-279 (2020) · Used in Functional hazard assessment with AI: ARP4761A, AC 25.1309
- Failure condition classification
The severity assigned to a failure condition in an FHA: catastrophic, hazardous, major, minor or no safety effect, judged by the effect on the aircraft, its occupants and the crew. Each class carries a qualitative and, for large aircraft, a quantitative probability objective under AC 25.1309-1B and AMC 25.1309. The classification is an engineering judgement that requires a written rationale; an AI tool may propose one and check it for consistency, but it does not decide it.
Source: FAA AC 25.1309-1B; EASA AMC 25.1309; EUROCAE ED-135 · Used in Functional hazard assessment with AI: ARP4761A, AC 25.1309
- SORA (Specific Operations Risk Assessment)
The JARUS methodology for assessing the risk of an unmanned aircraft operation in the specific category and deriving the safety objectives it must meet. It works in steps: describe the operation (ConOps), determine the ground risk class and the air risk class, apply mitigations, derive the Specific Assurance and Integrity Level (SAIL), and identify the operational safety objectives (OSO) for that SAIL. Version 2.5 is the current JARUS edition; EASA adopted it as AMC1 to Article 11 of Regulation (EU) 2019/947 by ED Decision 2025/018/R.
- SAIL (Specific Assurance and Integrity Level)
The level, from I to VI, that SORA assigns to an operation from its final ground risk class and residual air risk class. The SAIL sets the robustness with which each operational safety objective has to be met, from low to high, and it therefore sets how much evidence the operator must produce. It is the single number an authority looks for first in a specific-category application.
Source: JARUS SORA v2.5, Step 7
- ConOps (concept of operations)
The description of an intended operation in enough detail to assess its risk: the aircraft, the operational volume and its buffers, the ground and air environment, the crew, the procedures and the contingency and emergency responses. In SORA it is the first step, and everything that follows is only as good as it is. Military airworthiness frameworks use the term in the same way for the intended use of a system.
- Design organisation approval (DOA)
The approval that lets an organisation demonstrate its capability to design products, parts and appliances and to hold type certificates and approve certain changes itself. Point 21.A.14 sets out how the capability is demonstrated and Subpart J the requirements for the approval. A DOA holder's handbook and procedures, like a Part 145 exposition, are a controlled corpus an AI assistant can answer from.
Source: Regulation (EU) No 748/2012, point 21.A.14 and Subpart J · Used in Certification-basis compliance matrix: EASA Part 21 and FAA
- Approved corpus
The set of documents an organisation's administrators have ingested and approved as the only source an AI assistant may answer from: regulations at the agreed revision, expositions, procedures, standards and records. Everyday users can read from it and cannot add to it. When a question is not answered by the corpus, the assistant says so instead of answering from general knowledge.
Source: Can AI be trusted for aviation compliance documentation? · Used in Can AI be trusted for aviation compliance documentation?
- Clause-level citation
A reference from an answer to the specific paragraph of the specific document it came from, with a link that opens that paragraph. It is the unit of verification in compliance work: an answer with a clause-level citation can be checked in seconds by anyone, an answer without one has to be researched again. It is one of the four conditions under which AI output in compliance documentation can be relied on.
Source: Can AI be trusted for aviation compliance documentation? · Used in Can AI be trusted for aviation compliance documentation?
- Hallucination (in regulatory work)
An answer that is fluent and wrong: an invented paragraph number, a clause quoted from the wrong revision, an acceptable means of compliance presented as a rule, or an FAA requirement mixed into an EASA answer. These are the four failure modes that matter in compliance documentation. They are not fixed by a larger model; they are contained by restricting what the model answers from and by citing every statement so that a reviewer can check it.
Source: Can AI be trusted for aviation compliance documentation? · Used in Can AI be trusted for aviation compliance documentation?
EASA Part 21, FAA 14 CFR Part 21
EASA Part 21, FAA Order 8110.4C
EASA Part 21
EASA Part 21, FAA 14 CFR Part 21
EASA
EASA Part 21, FAA 14 CFR Part 21
EASA Part 145
EASA Part CAMO
EASA Part 145, FAA 14 CFR Part 145
EASA Part 145
EASA Part 145, Part CAMO, Part 21
EASA, Regulations (EU) 2023/203 and 2022/1645
EASA Part-IS
Deployment
Deployment
ED-135 / SAE ARP4761A, ED-279 for UAS
AC 25.1309-1B, AMC 25.1309, ED-135
JARUS SORA 2.5, EASA specific category
JARUS SORA 2.5
JARUS SORA 2.5, EASA specific category, defence
EASA Part 21 Subpart J
Wingman360 Teammate
Wingman360 Teammate
AI